04:45:48 UT, 2000 frames/second

Figure 5.11 shows the development of an angel sprite cluster at 04:45:48 UT. The sprite event was associated with an unusually energetic 165 kA +CG at 891 km range. The vertical bar in the figure shows the altitude in kilometers above the +CG. The error in height of the sprites would be $\sim\,$6 km for a 40 km radial displacement from the +CG location.

Figure 5.11: The correlated high-speed video sequence and charge moment change of the 4:45:48 UT sprite event. The horizontal lines are the same as before (see Figure 5.10 caption). The parent +CG's charge moment change exceeded the conventional breakdown threshold above $\simeq$75 km MSL altitude within the first frame and this produced diffuse luminosity consistent with the theoretical predictions of Pasko et al. (1997b). Angel sprites developed downward at unusually high velocity in frame $b$. The sprite cluster produced a large charge moment change in frames c-d. The charge moment curve was provided by Steve Cummer.

Figure 5.11 shows that the charge moment change of the parent discharge rose very rapidly. Within the initial millisecond, the charge moment exceeded the conventional breakdown threshold between the base of the ionosphere ($\simeq\,$81 km MSL) and $\simeq\,$75 km MSL altitude. This should have resulted in a diffuse luminous region due to large-scale ionization (referred to as a ``sprite halo'' in Barrington-Leigh et al. (2000)), as was indeed observed in frame $a$. The shape and altitude extent of the sprite halo is consistent with the theoretical predictions of Pasko et al. (1997b).

The dashed vertical lines in Figure 5.11 correspond to the light-integration periods of the second horizontal sensor block from the top of the high-speed imaging sensor (see Appendix A.1). This block was chosen because all but the lowest portion of the sprite halo was within this block. There was an apparent delay between the first appearance of the sprite halo, which occurred in frame $a$, and the onset of conventional breakdown at 80 km according to the calculated charge moment change. The apparent delay was due to the additional path length required for the electric field from the parent discharge to propagate upwards to produce the sprite halo before luminosity could propagate outwards to the observer along a similar path length taken by the sferic. This additional propagation path would result in emissions which were delayed by approximately the propagation speed of light up to 80 km altitude. This delay was 80/300$\,=\,$0.27 ms, which would result in conventional breakdown first appearing in frame a, as is observed.

The sprite halo reached a lower altitude of $\simeq\,$71 km MSL in frame $a$. However, by the end of frame $a$, the charge moment change had only exceeded the threshold at 76 km MSL altitude ($\simeq\,$700 C$\cdot$km) according to the calculations based on an ionosphere conductive ledge at 81 km MSL and the geometric considerations discussed above. The source of this discrepancy is partly due to the choice of the ledge altitude. The breakdown will greatly enhance the ambient conductivity and this will drop the altitude of the ledge. Conceptually, the electric field will be enhanced at a given altitude and charge moment change if the conducting ledge is lowered, since the distance to the images above the ionosphere will decrease. This will reduce the breakdown threshold at any given altitude. This demonstrates the importance of self-consistently modeling breakdown and relaxation-time parameters, as was shown by Pasko et al. (1997b).

Dropping the ledge from 81 km MSL altitude to just above 74 km MSL will reduce the altitude of breakdown shown in the charge moment curve for frame $a$ from $\simeq\,$76 km MSL to $\simeq\,$74 km MSL. If the center of the sprite halo was located at a range of $\sim\,$20 km beyond the +CG, then a feature at 74 km MSL altitude would have appeared to have been at 71 km MSL. However, the lower extent may also have been associated with local regions of enhanced breakdown which developed into streamers, as discussed next.

An enhanced luminous region which extended below the base of the sprite halo is shown by an arrow in frame $a$. A positive streamer developed (and split) downwards in frame $b$ from the enhanced luminous region at an average velocity of $3.4{\times}10^{7}$ m/s ($>$10% the velocity of light). The velocity of the positive streamers in the angel sprites were much faster than in carrot and columniform sprites (see Sections 5.2.3 and 5.2.4). The higher velocities are roughly consistent with the larger charge moment change and the theoretical model of tendril velocities put forth by Raizer et al. (1998). Several of the other angel sprites in frame $b$ were also preceeded by enhanced luminosity at the base of the sprite halo in frame $a$ and this is consistent with the development of angel sprites shown in Section 5.2.5.

A few of the tendrils of the angel sprites reached a lower terminal altitude of $\simeq\,$36 km MSL in frame $c$. Pasko et al. (2000) predicted that the tendrils would terminate at $\simeq\,$38 km under the influence of a 1000 C$\cdot$km parent discharge, which is roughly consistent with the cumulative charge moment change in frame $c$.

A second hump in the charge moment began in the latter part of frame $b$ and was likely correlated with current in the sprite, as discussed in Chapter 4. The magnitude of the sprite charge moment change is difficult to separate from the parent discharge charge moment since the latter may still have been varying. An extrapolation of the parent discharge slope at 1.2 ms reveals that the sprite charge moment may have been as large as 840 C$\cdot$km, which would be the largest charge moment change ever indicated for a nighttime sprite. As was shown in Section 4.3, daytime sprites can produce significantly larger charge moment changes than even this extreme event.